DINRG

• ~50 attendees per datatracker
• video recording
• chat log

DINRG Chairs’ Presentation: Status, Updates

• chair slides

Cyberspace Regulation Case Study

• Xing Li
• presentation slides 1
• interesting analysis about whether business leads regulation or vice
  versa (answer: yes, topic dependent)

• Dirk Kutscher In some juristicnions govs. are discussing
  centralization -- are you aware of any discussion in China about
  economic centralization and regulating decentralized.

  • Answer: Governments do get involved in regulationns
  • even in the beginning there were questionsn about whether and
    how best to connect to the internet
  • the great firewall being a good example result of that thinking

• Andrew Campling: One aspect of fragmentation is that TLS 1.3 is
  blocked (rumor)?

  • Answer: I don't know unfortunately
  • Geoff Huston (answering the TLS 1.3 question): Our ads are
    inside China.
  • we have no control over which TLS version gets selected.
  • I can't fingerprint what's behind the firewall unfortunately

• Geoff Huston: where we are is a direct result of frustration of the
  US in 1970s due to profits

  • result was deregulation of the private company in the US
  • theory was competition would get rid of profabilitiy
  • goal was intended to protect consumers
  • other economies were lately pressured in similar way to follow
    suit
  • competiton came in alternate systems, including the internet
  • the internet grew in a deregulated environment in a capatalistic
    system
  • everything above is not "Chinese" in the way things happen
  • My understanding within China is that there was more of a
    centralized system/rollout
  • Is regulation in China of a different form?
  • Big entities like chinamobile, etc have a "license" to operate
  • (end question)
  • Answer: Need to think about this as I haven't considered it
  • at the top level all the companies are doing the same thing
  • GH: in other places change in evolution comes from market
    changes

• Robert Carolina: following Geoff, AT&T was broken up by
  anti-competition law.

  • One result was people fighting for the right to connect without
    prior approval
  • the freedom to choose where to connect through helped develop
    the indepnedence of the internet
  • fragementation that should be avoided at all costs is technical
    fragmentation
  • especially name and address space
  • avoiding other types of fragmentation categories would be harder
    and maybe not as important
  • (no actual question, just a comment)

• Rudiger Volk:

  • listing very limited number of national ISPs
  • centralization happens at upper levels
  • understanding that space within china from the outside is
    difficult
  • what is focus of the research group? just layer 3?
  • The top level is much more valuable than the layer 3
    infrastructure, and muc more dangerous
  • chair's answer: all layers are important. The report lists all
    three areas as necessary to study.

• Tobias Fiebig:

  • technical fragmentation is dependent on all the other aspects of
    centralization
  • eg, other forms of centralization lead to technical
    centralization such as filtering/blocking by governemnts
  • (no question, just a comment)

Delivering Social and Municipal Services

• Wilfried Pinfold
• presentation slides 2

• list of social services available by state in the US

  • including coordination among them (or not)
  • need to address the challenges of being able to carry needed ID
    bits between services

• the need for multiple transportation apps for bikes/scooters/buses
  etc are an inconvenience for users

• building PASS (personalized access system for services) that fixes
  some of this pain with decentralization
• source code is online
• Q/A
• Dirk: can you talk more about the trust and security infrastructure?
  Is it PKI?
  • there are a number of standards we are trying to use
  • debate about pod-to-pod encryption
  • can you be sure who you're sharing information with?
  • do I need to protect the information in transmission?
  • looking at unique private keys to secure single transactions
  • this will be necessary to solve to solve real world problems
    safely
  • requires managing a large number of keys
  • definitely wanted a goal to keep it all open source

Workspace: A Local-First Application Prototype

• Xinyu Ma, Varun Patil, Tianyuan Yu
• presentation slides 3
• (follows a presentation at the last DINRG meeting)
• currently work requires connection to an application server which
  may not be possible even though a local connection may be possible
• device trustability in an open environment is a challenge
• identifiers should have exactly one semantic user

• workspace solution example

  • handles eventual consistency
  • uses NDN tesbed network
  • uses Sync NDN transport (vector state syncing)
  • had a new user spin up to develop an application in under a few
    weeks

• Question from chat: something about DDoS defense

  • that's a separate problem

• David Oran:

  • was referring to a computational attack against the nodes by
    doing syncronization verification
  • multi-way exchanges prevent this so that you don't just have to
    verify invalid signatures on the reception of any particular
    message
  • Lixia answer: NDN networks already helps this some.
  • take it to mailing list as its a more generic protocol

• Wes Hardaker: compare and constrast with git and a notification
  system

  • (I failed to get my question across because the answer was about
    NDN vs TCP/IP and not about the system comparison at the content
    layer functionality)

• Dirk Kutscher: how do you resolve overlapping inputs when offline

  • all changes will be accepted when users come back online
  • users will need to handle these conflicted changes

Discussion on RFC 9518 (“Centralization, Decentralization, and Internet Standards”)

• what can we do to move the work forward beyond this publication?
• first question: distributed vs decentralized definitions

• things centralized over the past decade: platforms, resources,
  popular applications, individual data

  • why?

• What should we aim to decentralize first?

  • Wes Hardaker: user data
  • Andrew Campling: protocols
  • Stephane B.: what about protocols? Are
  • Tobias: complexity is being enforced by centralization of
    platforms. Email is a good example that may require more and
    more elements of STS and DANE and etc. It forces STS to become
    the most popular because large operators set the state by
    refusing to take mail unless large organizations requiring their
    preferred solution.
  • Dirk K: it's more important to better understand the societal
    issues
  • Geoff H: we deregulated and accepted market forces. The drivers
    are investment. I'm only going to fund projects that completely
    dominate their environment. Economic driving forces require
    completely enclosing the user. Platforms are designed to resist
    competition. Deregulation caused this and thus
    size/volume/capturing-users matters. You need to first figure
    out how to drive the market away from its current desires.
    Investor return happens on massive scale and centrality. Layer 3
    has become unimportant compared to the application layer.
  • Lixia: that is important point, but the world is not that
    pessimistic.
  • Rahul: without a way to establish consensus there is no easy way
    to (audio cut-offs)
  • Vittorio: I'd much like to see messaging on the list. Email is a
    type but not all. Gathering of a lot of data is a problem.
    Increasing the complexity of the protocols is making everything
    more difficult to host services beyond the average person.
    Technical community has been promoting centralization and most
    of our community works for communities that promote this
    centralization.
  • Maryam: Big problem is non-existant taxonimy of what it means to
    be decentralized compliant.
  • Alexander: docker components can make mail servers easy to run,
    but your mail will still be rejected because you're not trusted
    and are being rejected by large players. (missed some)
  • Arnaud: I created an email platform early on. Very distributed.
    Email arriving provides a critical decision point that requires
    a decision. Can a protocol keep the conditions for when two
    worlds need to meet. VCs can be a person of innovation. How many
    of the layers in security make money? Most new things will fail.
    We miss the steps and nature of design from topolog, ethics,
    legal, etc.
  • Tobias: we are lost when complex systems (even in docker
    containers) fail.
  • Nicola: centralization is not just about protocol design but
    also markets and economics. EU has forced the app store to open
    up, eg. Regulations can help. GDPR has also helped, but at the
    same time has caused some companies to stop providing services
    in the EU.
  • Andrew C: (missed). Use of SDOs can help out manuever some
    elements.